AMMRC TR 72-9 IAD

SSTATIC AND CYCLIC FATIGUE

t, OF;CERAMIC MATERIALS

SAMU*L J. AC-QUAVIVA and RICHARD CHAITMATERIALS TEST ING DIVISION

March 1972

*Approved for putolb selam; diftribtion unliraited, M4AY 18 ~I

NATIONAL TECH4NICAL LIINFORMATION SERViCE C

ARMY MATERIALS AND MECHANICS RESEARCH CENTERWatertoww, A•,ssachusetth 02172

`7.T

Pm

SM

The findings in this report. are nvt. to he construed asar official Department of tht Army pozitton. unle~s sodesigna~ted by other authorized documents.

Mention of an~y trtdc names or wanufatture~s in thii ri~por,.shall oot be construtd' as adiertising nor as 3n official

- ~indorscement. or aprrov4il of such product~s or companies bythe United States Government.

DISPMSTIOr INSTRU--TIOL3

DeZttG t~is :Q10r% Ot*I. tI !3 2o in§fw: 9"(ded.to% rezarm it *10 Ibt Or~tdkaor.

UNCLASSIFIEDsvcu-8yV Classiftva-ion

DOCUMENT CONTROL D.* rA R & DfSrfc-efty etas &JfksaU.,2 of WIS. boj ot jAsftact SM Indevi, C nnotelaf im ust 1&e *suetel Whaft Ih, OV01*1I to p40t )a eaSS111dJ

OPGM- .lCIVT .. Ofpr~t&*U*G..) ZA. REPORT .ZICUPITV CIASSlWICAT8ON

Army Material's and Mechanics Research. Center Unclassi-fiedW-atertown, Massachusetts C2172 2~b.GoP

) tLPOAT T!8.CE

STATIC AND CYCLIC FATIGUE OF CERLAMIC K'A77ERIALS

4 `CFC%8PTkVE NOTVS (7Tj* af t~p~rt and inbCl8.sh dftst)

S,=ruel., J. AcquaviLva and Richard Chait

AS PORT CAIC 'U_ TOTAL NO0 011 P.CM3 #40 or -. fez

Marc~h 1972 10U S. CONTRACT 010 CRA14Y NO j8. ORtMIRA?0R1 RECOOOT NUatL142?

4. PROiLcy~ .0 /A 1F1622053AAS2 01 AINMAC TR 72-9

c-.k%0HS Cole 51 2E. I 1.A52-X031801 St. OTH4ER AEPOPIT KOISS (AeW *yc.A? Mnar. f. 09

4Agency Accession Number DA 0C4808 I_____________________Approved fo-r public release; distribution unlimited.

11.SUPPLECHCNARY NOTES "SffaWLV01 Arv-T

SU. So Arcy Materiel -0=andWashington, 0. C. 20315

85 OSRACTStatic and cyclic fatigue characteristics were deterguned for two celramic

materials, aa alumina AD-941 and a hot-pressed boron carbide B.C. The static fatiguelimits det~rmined in three-point flexure were 23,530 psi for A1203 and 43,000 psi forb.,C, while the cyclic fatigue limits deteimined by stressing a cnatilever-d beam wercI 1700 psi for the A1203 and 28,SOO p~.i for the Bi4C.

Examination or fracture surfaces by scanning and transmission electronmicroscopy revealed the fracture to bc )~f a transgranular and intcrgra~ular naturefor both matesijals. Fracture initiated at thc tension stwface ýGmd propagated immedl-azely to failure. There was nc change in fractuxe appearance in either the ststic orcyclic mode of failure. (Authors)

3ez1ity 1acuificstuon

UNCLASS IFIED

14 L'-1 d L..I, 6 LE*4K C

KtL ORO- -

~OL• U? ROL, S UT ROt.t 1 T

Ceramic materials IAluminaBoron c~rbide

Mechanical propertiesFatigueFracture toughness IMcicrostructureElectron microscopy.f

II

I I

__- _2 [ 11 I • __-- •" II _I -- IJI I

IUCLASSI:IESecurity Classicat!On

_ _• :,•.•.,•-• ,T• o -••.6••e. I i. .I..=•'.j

AM1 VA .: -0!

AMMRC TR 72-

STATIC AND CYCLIC FATIGUE OF CERAMIC MAT~f1dALS

Technical Report by

SAMNUEL J. ACJ121A VI VA and RICHA RD CHAIT

March 1972

D/A Project 1 F I62205AA52 01AMICMVS 0C-oi 512E1.11.A52-X031801Aircraft SurvivabilityAgency Accassion Number DA 0C4808

Anroved fat puiblic release. dis-mrution unlimited.

MATERIALS TESTING DIVISIONARMY MATERIALS AND MECHANICS RESEARCH CENTERWatertown, Massachusetts 02172

. R.' Wq"-AW17 .MVTTMIVVý! TIMM.,,

A[kMY MATERIALS AND MECHANICS RESEARCH CENTER

STATIC APP GYZLIC FATIGUE OF CERAMIC MATERIALS

AB3TRACT

Static and cyclic fatigue ,-haracteriýstics were determined for two ceramicmaterials, an alumina AD-94 and a hot-pressed boron carbide B4C. The staticfatigue limits determined in three-pint flexure were 23,500 psi for A120 3 and43,000 psi for B4C, while the cycii: 2atigue limits determined by stressing acantilevered beam were 21,700 psi fox ",h% A120 3 ard 28,500 psi for the B4C.

Examination of fracture surraces by scanning and transmission electronmicroscopy revealed the fracture to be of a transgranular and intezgranularnature for both materials. Fracture initiated at the tension surface and propa-gated immediately to failure. There was n,: change in fracture appearance ineither the static or cyclic mode of failure.

INTRODUCTION

There is increased emphasis on the structural use of armor material as evi-denced, for example, by the proposed Aerial Armored Reconnaissance Vehicle. 1 , 2

This vehicle utilizes monocoque construction with steel armor plate serving asboth the skin and primary structure. It is also possible that ceramic compositearmor, i.e., ceramic facing bonded to a suitable backup material may be consideredfor a nonprimary but critical application, especially under severe ballisticenvironment. However, add.iticnal information is needed regarding the behaviorof candidate materials that will serve in both a load-carrying and ballisticprotection capacity. A program is underway with the objective of characterizingcandidate materials. The present study covers aan initial phase of this charac-terization of ceramic composite armor by focusing on the static and cyclic fatigueresponse of two candidate ceramic materials, A120 3 and B4 C.

MATERIALS AND TEST PROCEDURES

Two materials were selected for investigation; (1) AD-94, a commerciallyavailable alumina A120 3 , manufactured by Coors Porcelain Company, Golden, Colorado;and (2) a hot-pressed boron carbide B4C, manufactured by the Norton Company,Worcester, Massachusetts. -the microstructures of these mmterials are shown 4nFigure 1, and their properties are given in the. following table:

Table I. Physical Pr3perties

Material Flexural Strength Modulus Density !Grain Size

A120 3 38,000 psi 41.5 x 106 psi 3.71 g/cc IOU

B4C 51,000 psi_ 60 x 106 psi 2.43 g/cc 10uj

Ib. .9 41.le

9,A

a. Alumin3 94% b. Boron Carbide

Figure 1. PHOTOMICROGRAPHS OF MATFRIALS T.STED. Mdg. 50OX

The static fatigue tests "-tre conducted in three-point bending with a dead-weight load suspended from the center of the specimen. Tha specinen size forboth the static and cyclic fatigue tests w.as 1/4" x 1/4" x 3Y with a surfacefinish of lo tis.

For purposes of this study, cyclic fatigue has been defined as an inter-mittently applied load to a specimen such that the stress on the surface of thespecimen follows a sinusoldal excursion varying from zero to a Maxim,= tensileor compressive stress, depending on the applied load. The dynamic test consiszedof applying a load at the end of a cantilevered specimen. at the rate of 15 cps.The equipment utilized in this test has been describcd elsewhere. 3

The arbitrary cut-off limit for the cyclic test was 1 0 cycles in r1.eevent failure did not occur. Several tests beyond this arbitrary lildit wereconducted; these tests indicated that a limit of 1 x 106 cycles was justified.

RESULTS AND DISCUSSION

The flexural strength of the A12 0 3 w-.en )oaded to failure, was 38,000 psi;th'.s will be referred to as the reference strength, Figure 2. At 7S% of th•reference strength life was limited to !0 seconds. Decreasing the stress levelto 70% or 27,000 psi increased the static fatigue life to 13-1/2 hours. At 60_of the reference strength or 23,000 psi, life was extended to 72 hours, at whichtime the test has terminated, it being assumed that life would be infznite atthis stress level. The specimens that were not fractured in the static testswere rtloaded to failure, with failure occuiring at 35,000 psi or Within 6 or 714

of the reference strength, indicatingthat virtually no damage was incurred

5O within the test specimen during the stati.•61 min B4test. It would thus appear that this

-material can be loaded statically at 60%of the reference levei for an indefinite

40- period of time.

mStatic

fatigue tests on BnC were

4 X conducted undei the same atmospheric con-

a. ditions as for the A1 2 0 3 , namely 72 F30 10 sec and 35% R.H. The reference strength of

_13% hr the B4C was determined to be 51,000 psi.��94AD._ At 90% of this value or 46,000 psi, frac-

ture occurred after 61 minutes; at 75%X ~and 73% of the refe--ence strength, static

fatigue life exceeded the arbitrary limitof 72 hours. The specimens that did notfracture in the static tests were re-

20 60 ell loaded, and in each case fracture oc-Life - hours curved at 97% of the reference strength,

Figure 2. STATIC FATIGUE OF CERAMIC indicating once again that static fatigueMATERIALS had no detriuentai effect on t.,_ z_,ength

of the material.

A.V

One of the more predaominant mech~anisms leading to frac-ture is stres;s corro-sion,* and it is felt fur'ther tests at humidities az~proaching 100%'1 and also testsconducted in a vacuum would el*;cidate the rGle of stress corrosion in fatigue.Also, additional tests to de~e;`:iteiv establish the observed trend are planned.

Several techniques hav- been reported for increasing the strengti. o~ ceramicmaterials. Semple-,' for example, introduce4 a compressive .tress on the surfacerf Al2O pcmn yaha treat process, and subsequent leading of the heat-

treated specimens showed an iscreasle in strength over the tuti eated specimens.Kirenner ez al.,- recognizing that tailures occur at surface Iflaws, treated thesurfaces by refirin~g, cheaic-4 and flame polishing, cheatical etching, and glaz~ngThese treazr.ents also proved to be beneficial in increasing the strength of th.-treated material. The effect that these treatm'ents might have on improving thefatigue: life of the A120-. and B4C should also be investigated.

The results obtained for the cyclic 'testing of the A4120 3 and B,.C are shownin Figure 3. The reference strengths for each material are the same as in thestati-c tests. The threshol1d level is defined as that level of stress below whichan infinite fatigue2 life is anticipated. The threshold level for both the AlOw:and R1.C in cyclic. fatigue was determin2d to be at 5`0 of thc reference stren-gthfor each material. The- data for A1201 compares favorably with that reported byWilliams6 where he reports a "cyclic endurance limit of 0.56 of the single stroke

Scanning electron microscopy (SEM%) of the f;.-acture surface, Figure 4, indi-cates fracture is of both a tranisgranular anti i'ttergraniular nature. The locationof the area shown in the micrograph is approximai.ely at the center of the frac-ture surface. Transmixsion electron microscory of the fracture surface of Al-O-.,

50k

-' 0

Cycles&.gre.3FLXUALFAIGU Q CRAICMA&RA4

xOsuso 3ih0 .tos iws Rsdc ntttKna Cit.Mssui

O_ No ractur

44

c Al~ 2 A. 0M310 1;OO d B C Mttj 1525X

F:Spre 4. FRACTURRE SURFACES BY SEM19-066&I4511A.MC.71

Figure 5 indicates th-' the tran!.g:anular fracture involves cleavage. It Wotildalso appe~ar tla,,t the.- are some c~cý,vaesessoni h E irgah~

Itho BL.C, Figpre 4. e-_-sessonIth Emirgahf

Fracture mechaznics concepts, based upon the Griffith criterion, w-ere u!!edto determine ;5train energy release rate andJ ceitical flaw size for the twc% mate-rials. The strain energy release rate was determined by fracturing a V-rnotche6specimen, 1/14"x 1/411 2-1/2", having a notch dzpth of O.OSG inch. The speci-mens were fractured in three-point bending as outlined by iDavidgr- and Tappin'who defilned the strain energy release rate G by

4.4

Figure 5. FRACTURE SURFACE OF A120 3

BY TEM (CARBON REPLICA) MAag. 15,000X

4 G -= (-2). Gr2c 12E

where . is Poisson's ratio; r is Young's modulus in pounds per square inch; VFis the fracture stress in pounds per square inch; and c is the notch depth ininches.

Notchoc specimens of both Al.03 and BtC were fractured in an Instron test-ing machine at a crosshLad speed of 0.020 inch per minute Substituting the testdata into Equation 1, the strain energy release rate for A120I was 0.270 in-lb/sq in. (4.73 - 10' erg/cm2) and for the B4C, 0.192 in-lb/sq in. (3.36 v 104 ergs/m2). For the A12 03 , the strain energy release rate compares favorably with

0.302 in-lb/sq in. reported by McKinney.ý Care should be exercised in comparingfracture energy data, since Gross and Gutshal!9 found that the strain energy re-lease rate was grain-sFze dependent.

From the Griffith criterion

Gc = G2 ',a/E. (2)

Solving Eqtuation 2 for a, a = (GcE)/_o 2 . If a is assumed to be the fracturestress of -n unnotched bar, then it is possible to estimate the critical flawsize from Equatioa 2. From the test data, it was found that the critical flawsize for A1203 was 0.003 inch, approximately 7Sum or about 8 grain diainetern;flaws of this size should be readily detected by nondestructive techniques. Thecritical flaw size for B4C was found to be 0.0015 in.•, approximately 4 0-im, orabout 4 grain diameters. It should be pointed out that flaws of this naturewere not ob.erved in the examination of the fracture surfaces by scanning elec-tron microscopy. Additiono' tests are planned in the area of f-acture energymeasurements.

4' 5

SUMMARY

The static and cyclic fatigue properties of 01o materials, an alumina AD-94and a hot-pressed boron carbide B4C, have been studied. The static fatiguest-rength of the A120 3 at 72 hours was 60% of the reference strength, and for theB4C the static fatigue for 72 hours was 74% of the reference strength. Staticfatigue was determined by applying a dead-weight load to the center of a specimei.of 1/4-" - I/41 x 3" dimensions; the specimen was supported at a span length of2-1/2 inchess.

The cyclic fatigue properties were determined by applying a load to the eadof a cantilevered sj~cimen. The specimen size correspondtd to that used forstatic fatigue, with the cantilevered span being 2-1/2 inches; this ensured thestressed volume being the same for both the static and cyclic fatigue testz.The threshold level for safe cyclic fatigue contitions for both A120 3 and B4Cappears to be at 57% of the reference strength level. Examination of the frac-ture surface revealed the fracture to be of a transgranu-ar and intergranularnature, with the fracture initiating at the tension surface; ho;evex, no evidenccwas present of intermittent progressive crack propagation. This latter observa-tien would indicate that once the crack was initiated, propagation wasinstantaneous.

From fraciutie mechanics concepts it wa.• calculated thac the scrain energyrelease rate was 0.270 in-lb/so in. for A120 3 and 0.192 in-lb/sq in. ior BLC,and the criti. :al flaw size was datermined to be 75.rn for A1203 and 40,. iLor BLC.

46

-5

34°

LITERATURE CITED

1. AAF7b, ior!: AiArcrqft,, Vert-iflight, v. 16, December 19710,

2. WONG, A. K.,* und BERMAN, 1. iqih z~cA--r-r-r - A P?_vie-w. As-myMateri-Als aad Machanilcs Research Center, A~M-RC HS 7141, April 19711.

3. ACQUAVIVA, S. J. 2 ._'!etkxd jfor Fc-igue T-iing Mazr.~Yteri~als. ReviewOf S-ientific instiumnents. rncember 1971.

4. SENPIZ, C. W. _1 S.'-es Deter i'nation ir' A!-a'P;A Bodes. -k.4ay Matz.-riais and Mechan~ics R#'-search Center. AukthRC TR 70-1 ., Augt~ist 190.

S. KiRCr-LiR, H. P. , CRUVER, R. M.*, and WALKZER, it. E. :.eqhEffect es .-irng fron: 2it urfcre T-eaz=en-t. PrTesented at Symposiur, on -the Scienceof Ceramic Machining and Surfacc Finishing, X-ES, Novembher 2-4, 197i). Tobe prublished.

6. WILLIAM.S, L. S. Ftique zn-e-rr-ic in o4cnc ~e~e f EngirseeI_ý-ing C i~s. W. tt'orth Kriegel. and Hayne Pialniour '11, ed., Interscience

P Aicatiop, New "ork - London. 1960, p. 245-A02.

7. DAVIDGE; R. W., adm -&AP!KN, G. Zh .:fe.,tive Furface Energy o' By-,ttle.MJa-ezia1e. Journal :uf MuzeTial Scictnce~v. 3, 196~.3 0'. 165-173.

X cKZINNEY. K. R. :'ractlure 7cug~vnez o --' ercr.:c Mateu-ials *Ysing the .5igzau-i-j~-Nac 3rs caimen. NRL Mz mrarsiuu Report. 1931, September 1968.

n ROS5, G,. F-., and QrSAL P. 1. .'a~.E ,.ikr-nj Measuent in5;-ciaCeyn;-cj crd Dlesian. S. J. Acquaviva al~d S. A. iortz, ech, Gord:r. andBreach, New York, 1969. p. 149-163.